[PATCH] ALSA: use __devexit_p

[PATCH] ALSA: use __devexit_p

[Randy Dunlap]

From: Randy Dunlap <randy.dunlap@oracle.com>

Change __devexit to __devexit_p:
sound/isa/opl3sa2.c:956: error: expected expression before '__attribute__'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
---
 sound/isa/opl3sa2.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- linux-2.6.22-rc5-git8.orig/sound/isa/opl3sa2.c
+++ linux-2.6.22-rc5-git8/sound/isa/opl3sa2.c
@@ -953,7 +953,7 @@ static int snd_opl3sa2_isa_resume(struct
 static struct isa_driver snd_opl3sa2_isa_driver = {
 	.match		= snd_opl3sa2_isa_match,
 	.probe		= snd_opl3sa2_isa_probe,
-	.remove		= __devexit( snd_opl3sa2_isa_remove),
+	.remove		= __devexit_p(snd_opl3sa2_isa_remove),
 #ifdef CONFIG_PM
 	.suspend	= snd_opl3sa2_isa_suspend,
 	.resume		= snd_opl3sa2_isa_resume,

Re: [PATCH] ALSA: use __devexit_p

[Jeff Garzik]

This reminds me...

Someone needs to go through ALSA and audit all delays executed via 
$FOO_interruptible().

Several delays within ALSA wait for hardware conditions, and do not 
check for signals pending, which means that the wait-for-condition loop 
becomes a busy loop:

	while (1) {
		foo = read_hardware()
		if (foo & interesting_bits)
			break;
		schedule_timeout_interruptible(1);
	}

The proper fix is (a) remove "_interruptible" [recommended] or (b) check 
for signals.

grep'ing for "_interruptible" quickly finds several such ALSA bugs.

	Jeff

Re: [PATCH] ALSA: use __devexit_p

[Takashi Iwai]

At Sat, 23 Jun 2007 14:40:22 -0700,
Randy Dunlap wrote:
> 
> From: Randy Dunlap <randy.dunlap@oracle.com>
> 
> Change __devexit to __devexit_p:
> sound/isa/opl3sa2.c:956: error: expected expression before '__attribute__'
> 
> Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>

Thanks, I applied them to ALSA tree now.


Takashi

Re: [PATCH] ALSA: use __devexit_p

[Takashi Iwai]

At Sat, 23 Jun 2007 18:50:57 -0400,
Jeff Garzik wrote:
> 
> This reminds me...
> 
> Someone needs to go through ALSA and audit all delays executed via 
> $FOO_interruptible().
> 
> Several delays within ALSA wait for hardware conditions, and do not 
> check for signals pending, which means that the wait-for-condition loop 
> becomes a busy loop:
> 
> 	while (1) {
> 		foo = read_hardware()
> 		if (foo & interesting_bits)
> 			break;
> 		schedule_timeout_interruptible(1);
> 	}
> 
> The proper fix is (a) remove "_interruptible" [recommended] or (b) check 
> for signals.
> 
> grep'ing for "_interruptible" quickly finds several such ALSA bugs.

Yep...  I fixed now on ALSA tree.  Thanks for hints!


Takashi